Enclosure experiment on a eutrophic lake: additive or synergistic?
A shallow lake (mean depth 1.8 m) has had cyanobacterial blooms for a decade. Managers run a 12-week summer enclosure experiment: 12 enclosures of 2 m diameter, open to the sediment, three replicates per treatment in a 2 × 2 design.
Treatments: N = 50% reduction of external phosphorus load, achieved with a phosphorus-binding agent; F = exclusion of fish longer than 5 cm by netting, allowing large Daphnia to establish; C = control; NF = both.
End-of-experiment means (replicate SE under 10% of each mean):
- C: total P 92 µg/L, chlorophyll a 48 µg/L, Daphnia 8 ind/L, Secchi 0.4 m
- N: total P 61, chlorophyll a 39, Daphnia 11, Secchi 0.5 m
- F: total P 88, chlorophyll a 21, Daphnia 65, Secchi 0.9 m
- NF: total P 58, chlorophyll a 8, Daphnia 86, Secchi 1.7 m
Sediment phosphorus release measured in intact cores from each enclosure averaged 4.1, 3.8, 3.6 and 2.2 mg P m⁻² d⁻¹ for C, N, F and NF respectively.
1. Compare the treatments. Is the chlorophyll response to combined treatment additive or interactive? Show the arithmetic.
Read the worked answer
Start with the proportional effects.
- Nutrient reduction alone: 39 ÷ 48 = 0.81, a 19% reduction.
- Fish exclusion alone: 21 ÷ 48 = 0.44, a 56% reduction.
- If the two acted independently and multiplicatively, NF would be 48 × 0.81 × 0.44 = 17.1 µg/L.
- An additive model on absolute reductions gives 48 − 9 − 27 = 12 µg/L.
- Observed NF is 8 µg/L, below both predictions, and 53% below the multiplicative expectation.
So the effect is synergistic, not additive. The mechanism is visible in the other columns. Fish exclusion works through grazing: Daphnia rise from 8 to 65 ind/L and suppress algal biomass, and chlorophyll falls even though total phosphorus stays at 88 µg/L. That decoupling is important, because it means total phosphorus alone does not predict chlorophyll across these treatments. Under combined treatment the water clears (Secchi 0.4 to 1.7 m), which lets light reach the sediment and can shift the lake toward rooted macrophytes, stabilising the sediment and reducing release: core release falls from 4.1 to 2.2 mg P m⁻² d⁻¹ in NF only.
The weakness of nutrient reduction alone is also visible. Total phosphorus falls 34% to 61 µg/L but chlorophyll falls only 19%, because internal loading from the sediment buffers the water column at a concentration still sufficient to support a bloom.
2. A funding agency argues that nutrient reduction alone is cheaper and "addresses the root cause". Construct the strongest evidence-based response.
Read the worked answer
Both claims contain part of the truth, and the response should separate them rather than pick a winner.
The agency is right about durability. External load reduction is the only intervention that eventually lowers sediment phosphorus and therefore the only one likely to be permanent. But the experiment shows it is not sufficient on the timescale decision-makers care about. In 12 weeks, total phosphorus fell to 61 µg/L - still above the roughly 30–50 µg/L range where shallow-lake blooms are typically suppressed - and chlorophyll barely moved. Sediment release in the N enclosures (3.8) was essentially unchanged from controls (4.1).
The honest caveat is that a 12-week trial cannot detect the multi-year decline in sediment phosphorus that follows external load reduction, so the experiment should not be used to claim nutrient reduction fails. It shows nutrient reduction is slow.
Fish exclusion is the mirror image: fast and large, but plausibly reversible. If the fish community re-establishes and external nutrients remain high, Daphnia are grazed down and the algae return. Predictions are therefore: nutrients only, slow and partial; fish only, rapid and fragile; both, rapid and possibly self-sustaining because the clear-water state reinforces itself. The defensible recommendation is combined treatment with long-term nutrient control as the backbone.
3. What are the limits of scaling these enclosure results to the whole lake, and what design would you recommend instead?
Read the worked answer
Enclosures of 2 m isolate a water column that the lake does not isolate. They exclude larger fish and piscivores, eliminate wind-driven mixing and horizontal transport, and impose wall effects on light, turbulence and zooplankton behaviour. The sediment is enclosed too, so any lateral groundwater or sediment phosphorus input is cut off. Twelve weeks covers one summer and misses autumn turnover, winter die-off and the next spring's recruitment - precisely the periods that determine whether a clear-water state persists.
Statistically, three replicates per treatment constitute pseudoreplication if the inference is about the lake: the true unit of replication is one lake, and four treatments in one water body cannot estimate lake-to-lake variability. Secchi measured in a 2 m tube also overstates clarity relative to an open water column.
The recommendation is a whole-lake design, ideally a before-after-control-impact comparison across paired lakes matched for depth, phosphorus loading and fish community, with at least two treatment and two reference lakes monitored for several years. If whole-lake work is impossible, keep the enclosures as a mechanistic screen, add a mesocosm scale, and monitor fish community structure, sediment phosphorus fractions and macrophyte cover alongside chlorophyll. Also note that the phosphorus-binding agent itself may have altered sediment chemistry, so the NF sediment-release result partly reflects the additive as much as the biology.